大语言模型中文化理解的神经层面分析
Neuron-Level Analysis of Cultural Understanding in Large Language Models
AI总结：
研究通过神经层面分析揭示大语言模型中文化理解的机制，发现文化通用与特定神经元对文化理解有重要影响，且训练数据影响其文化理解能力。
摘要：
随着大语言模型（LLMs）在全球范围内日益广泛应用，确保其公平和全面的文化理解变得重要。然而，LLMs表现出文化偏见和对少数文化意识有限，而其文化理解的机制仍未被深入研究。为填补这一空白，我们进行了神经层面分析，以识别驱动文化行为的神经元，引入基于梯度的评分方法并添加过滤器以实现精确细化。我们识别出文化通用神经元，无论文化如何都对文化理解有贡献，以及与特定文化相关的神经元。文化通用和特定神经元仅占所有神经元的不到1%，集中在浅层到中层MLP层。我们通过展示抑制它们会显著降低文化基准测试的表现（最高降低30%）来验证其作用，而对一般自然语言理解（NLU）基准测试的表现影响较小。此外，我们表明文化特定神经元不仅支持目标文化的知识，还支持相关文化的知识。最后，我们证明在NLU基准测试上训练可以减少模型的文化理解能力，当更新包含许多文化通用神经元的模块时。这些发现为LLMs的内部机制提供了见解，并为模型训练和工程提供了实际指导。我们的代码可在https://github.com/ynklab/CULNIG获取。
英文摘要：
As large language models (LLMs) are increasingly deployed worldwide, ensuring their fair and comprehensive cultural understanding is important. However, LLMs exhibit cultural bias and limited awareness of underrepresented cultures, while the mechanisms underlying their cultural understanding remain underexplored. To fill this gap, we conduct a neuron-level analysis to identify neurons that drive cultural behavior, introducing a gradient-based scoring method with additional filtering for precise refinement. We identify culture-general neurons contributing to cultural understanding regardless of cultures, and culture-specific neurons tied to an individual culture. Culture-general and culture-specific neurons account for less than 1% of all neurons and are concentrated in shallow to middle MLP layers. We validate their role by showing that suppressing them substantially degrades performance on cultural benchmarks (by up to 30%), while performance on general natural language understanding (NLU) benchmarks remains largely unaffected. Moreover, we show that culture-specific neurons support knowledge of not only the target culture, but also related cultures. Finally, we demonstrate that training on NLU benchmarks can diminish models' cultural understanding when we update modules containing many culture-general neurons. These findings provide insights into the internal mechanisms of LLMs and offer practical guidance for model training and engineering. Our code is available at https://github.com/ynklab/CULNIG
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
